FunctionsVector
Vector
6 Vector functions in the NYXDB native function registry, with signatures, briefs, and measured medians where available.
6 functions in the vector family. Each name links to its page; … marks the variadic argument list (the engine declares no named parameters). A Measured value appears only where a benchmark corpus case joins by name (server-class hardware, median).
| Function | Type | Description | Measured |
|---|---|---|---|
cosine_distance(…) → Float64 | scalar | Cosine distance (1 - cosine similarity) of two numeric vectors. | — |
cosine_similarity(…) → Float64 · alias cosine_distance | scalar | Cosine distance (1 - cosine similarity) of two numeric vectors. | — |
dot_product(…) → Float64 · alias inner_product | scalar | Inner (dot) product of two numeric vectors. | — |
inner_product(…) → Float64 | scalar | Inner (dot) product of two numeric vectors. | — |
l2_distance(…) → Float64 | scalar | Euclidean (L2) distance between two numeric vectors. | — |
negative_inner_product(…) → Float64 | scalar | Negative inner product (maximum-inner-product ANN ordering form). | — |
to_valid_utf8
Repairs invalid UTF-8 byte sequences by replacing malformed segments with '?', returning valid UTF-8 text.
cosine_distance
Computes 1 - dot(a, b) / (||a|| * ||b||) over two equal-length array(float64) embeddings encoded as JSON-array text. Range [0, 2]: 0 = identical direction, 1 = orthogonal, 2 = opposite. Length mism…